But while distinguishing properties from capacities may help us to correctly interpret the quote above, there is another problem with it that is not so easy to solve: the fact that Deleuze and Guattari seem to be using two incompatible definitions of the term “assemblage”. In his own texts, Deleuze uses the term to refer not only to social assemblages, like the man-horseweapon assemblage, but to biological ones (the wasp-orchid symbiotic assemblage) and even non-organic ones, like the assemblage formed by copper and tin when they interact to form an alloy, bronze, with its own emergent properties and capacities. [Nota de rodapé 5: Gilles Deleuze and Claire Parnet. Dialogues II. (New York: Columbia University Press, 2002), p. 69.] But in his joint work with Guattari the term refers only to social assemblages. Thus, an army is both a “machinic assemblage” or an intermingling of material bodies (human, animal, technical bodies) as well as a “collective assemblage of enunciation”, that is, a whole in which statements have the capacity to create social obligations, like the commands that flow downwards in an army hierarchy, or the reports that flow upwards. [Nota de rodapé 6: Gilles Deleuze and Félix Guattari. A Thousand Plateaus. Op. Cit. p. 88.] In this second sense, the term “assemblage” is not only restricted to social wholes, since only in them can statements be speech acts, but it applies to only one level of scale: a different term is used to refer to the components of an assemblage, “bodies”, and another term to refer to the larger social wholes that assemblages form, “the social field” or “the socius”. Thus, whereas in the first definition the components of a  an-horse-weapon assemblage can themselves be viewed as assemblages, in the second one they cannot. (DeLanda 2010:72-3)
DELANDA, Manuel. 2010. Deleuze: History and Science. New York: Atropos Press.